sata-vsc broken on SGI Prism

sata-vsc broken on SGI Prism

[Andreas Schwab]

I'm getting this panic when loading sata-vsc on a SGI Prism:

Loading sata_vsc
ACPI: Unable to derive IRQ for device 0001:00:03.0
ACPI: PCI Interrupt 0001:00:03.0[A]: no GSI
FRZ XN request short error. 
FRZ XN error detail 1A:
	Command word: 0x11
	Supplemental: 0xff
	Source node id: 0x0
	Source chip id: 0x2
	Destination chip id: 0x0
	Data error: 0x0
	Echo: 0x70
	Valid: 0x0
	Address: 0x1802000021
FRZ XN error detail 1B:
	Command word: 0x0
	Supplemental: 0x0
	Source node id: 0x0
	Source chip id: 0x0
	Destination chip id: 0x0
	Echo: 0x0
	Valid: 0x0
	Address: 0x0
Fatal ICE HW error
Kernel panic - not syncing: hub_eint_handler: Fatal TIO Error

Bisection has identified this patch (together with the follow-up patch to
fix the compilation failure):

commit 7cbaa86b937b0b1fab95c159989f6a3c00bbcf78
Author: Dan Wolstenholme <daniel@wolstenholme.net>
Date:   Tue Jan 9 05:59:21 2007 -0500

    [libata] sata_vsc: support PCI MSI
    
    Signed-off-by: Jeff Garzik <jeff@garzik.org>

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: sata-vsc broken on SGI Prism

[Andrew Morton]

On Sat, 24 Mar 2007 10:17:24 +0100 Andreas Schwab <schwab@suse.de> wrote:

> I'm getting this panic when loading sata-vsc on a SGI Prism:

With what kernel?

> Loading sata_vsc
> ACPI: Unable to derive IRQ for device 0001:00:03.0
> ACPI: PCI Interrupt 0001:00:03.0[A]: no GSI
> FRZ XN request short error. 
> FRZ XN error detail 1A:
> 	Command word: 0x11
> 	Supplemental: 0xff
> 	Source node id: 0x0
> 	Source chip id: 0x2
> 	Destination chip id: 0x0
> 	Data error: 0x0
> 	Echo: 0x70
> 	Valid: 0x0
> 	Address: 0x1802000021
> FRZ XN error detail 1B:
> 	Command word: 0x0
> 	Supplemental: 0x0
> 	Source node id: 0x0
> 	Source chip id: 0x0
> 	Destination chip id: 0x0
> 	Echo: 0x0
> 	Valid: 0x0
> 	Address: 0x0
> Fatal ICE HW error
> Kernel panic - not syncing: hub_eint_handler: Fatal TIO Error
> 
> Bisection has identified this patch (together with the follow-up patch to
> fix the compilation failure):
> 
> commit 7cbaa86b937b0b1fab95c159989f6a3c00bbcf78
> Author: Dan Wolstenholme <daniel@wolstenholme.net>
> Date:   Tue Jan 9 05:59:21 2007 -0500
> 
>     [libata] sata_vsc: support PCI MSI
>     
>     Signed-off-by: Jeff Garzik <jeff@garzik.org>
> 

It seems that little of that patch remains present in current mainline.

Re: sata-vsc broken on SGI Prism

[Andreas Schwab]

Andrew Morton <akpm@linux-foundation.org> writes:

> On Sat, 24 Mar 2007 10:17:24 +0100 Andreas Schwab <schwab@suse.de> wrote:
>
>> I'm getting this panic when loading sata-vsc on a SGI Prism:
>
> With what kernel?

Linus' latest at this point in time.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: sata-vsc broken on SGI Prism

[Andreas Schwab]

Andrew Morton <akpm@linux-foundation.org> writes:

> On Sat, 24 Mar 2007 10:17:24 +0100 Andreas Schwab <schwab@suse.de> wrote:
>
>> I'm getting this panic when loading sata-vsc on a SGI Prism:
>
> With what kernel?

It's still the same with 2.6.21-rc5.

>> Bisection has identified this patch (together with the follow-up patch to
>> fix the compilation failure):
>> 
>> commit 7cbaa86b937b0b1fab95c159989f6a3c00bbcf78
>> Author: Dan Wolstenholme <daniel@wolstenholme.net>
>> Date:   Tue Jan 9 05:59:21 2007 -0500
>> 
>>     [libata] sata_vsc: support PCI MSI
>>     
>>     Signed-off-by: Jeff Garzik <jeff@garzik.org>
>> 
>
> It seems that little of that patch remains present in current mainline.

I'd guess the real bug is in the SN code.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: sata-vsc broken on SGI Prism

[Jeff Garzik]

Andrew Morton wrote:
> On Sat, 24 Mar 2007 10:17:24 +0100 Andreas Schwab <schwab@suse.de> wrote:
>> commit 7cbaa86b937b0b1fab95c159989f6a3c00bbcf78
>> Author: Dan Wolstenholme <daniel@wolstenholme.net>
>> Date:   Tue Jan 9 05:59:21 2007 -0500
>>
>>     [libata] sata_vsc: support PCI MSI
>>     
>>     Signed-off-by: Jeff Garzik <jeff@garzik.org>
>>
> 
> It seems that little of that patch remains present in current mainline.


An unfortunately illusion of further changes.  The "turn MSI on" stuff 
is still here, but Tejun's cleanups made some of that housekeeping code 
go away.

But, as Andreas was guessing, the problem here is the platform's MSI 
support, with this as an unfortunate symptom.

	Jeff